annsa
Log inTry Annsa
What's new

Changelog

What we've shipped, week by week.

Specs that know your codebase.

Connect GitHub and every spec uses your real files and naming. Each spec gets a number you can say out loud, and annsa can answer how-to questions about itself inside your editor.

Delivery

Connect your repo — every spec uses it

Connect GitHub and specs come back with your real file paths and naming, so you can build straight away — and annsa re-grounds the specs you already have the moment you connect.

Delivery

Every spec has a number — Spec #14

Each spec carries a short number wherever it appears — in the panel, the export, and circuit.spec — so everyone knows which spec you mean.

Intelligence

Ask annsa how it works, in your editor

A new circuit.help tool answers how-to questions like "how does scoring work?" from Cursor and Claude Code — so you get the answer without leaving your build.

Discovery

Rank and context, side by side

The brief shows what's driving a priority and the customers behind it as two clear cards that travel into the export — so the whole story copies out in one go.

Also this weekAsk sticks to the customers it can see, the free-tier rollover runs on schedule, the default project no longer counts against your plan, and the login screen carries a fresh line.

See if what you shipped worked.

When you ship, annsa asks your customer whether it resolved what they raised — and records the answer. Every brief now names the one metric it's meant to move. And your Radar briefing answers back when you reply in Slack.

Delivery

Share Back asks if it resolved the request

When a spec ships, the customer's branded page asks Yes or Not yet — so you know whether it landed, not just that you sent it. Say Not yet and annsa captures what's still missing.

Intelligence

Every brief names the metric it should move

Each spec states the one metric it's meant to move, and where to read it — so you know what success looks like before you build.

Intelligence

Reply to your Radar in Slack

Reply to the Radar briefing in its Slack thread and annsa answers right there — so a follow-up never leaves the channel.

Delivery

Connect your editor in a couple of clicks

New connection screens in the app and CLI hand you the exact prompt to copy — so annsa's priorities and specs reach Cursor and Claude Code fast.

Also this weekYour product switcher is on for every account, priority rows show how many customers are behind them, and long feedback quotes stay tidy in a brief.

Close every loop in your brand.

Share Back — the loop that runs to your customer, not to you — is now fully yours. Set your sender identity once: your company name, your logo, your colour. A live preview shows both the email and the in-app widget exactly as they'll land, so you approve the real thing before a single message goes out.

Delivery

Share Back, in your brand and previewed

The old Notifications tab is now Share Back — named for the loop it closes. Set your sender identity once: company name, from-name, logo, accent colour, footer. A live preview shows both the email and the in-app widget exactly as they'll land, so you approve the real thing, in your voice, quoting what the customer sent.

Also this weekBilling now reads "Feedback processed this billing cycle" with the reset date shown; a customer drilldown keeps its place in the URL on refresh; and the feedback widget button and brand colour pickers carry proper labels for screen readers.

Find any question you've asked annsa. Onboarding that learns your product.

Ask annsa is now a place you navigate, not a feed you scroll — every thread carries the spec, customer or private chat it belongs to, and a filter row finds the one you're after. Onboarding learns your product on the way in: its name, its stage, what matters most right now. And feedback that arrived in another language reads in the words the customer wrote.

Intelligence

Ask annsa — find the thread you're looking for

Every conversation now wears a chip for what it's grounded in — Spec, Customer, or Private — and a filter row above the list narrows to any of them, or to Archived. Rename a thread from its row menu and the list updates the moment you do; an archived thread is one click away under Archived, and one more to restore. Answers render with real headings and quoted customer lines, so a long reply reads like a memo, not a wall of text.

AskNew conversation
What did we deprioritise last month, and why?
Three priorities dropped out of Next in April. The biggest was Version history — dismissed on Apr 4 after two enterprise accounts churned and the evidence score fell below threshold. The full reasoning opens beside this chat.
Has it come back since?
Yes — 5 new signals this week, across 2 accounts. Radar flagged it under Coming back on Monday.
Ask anything…
Enter to send · Shift+Enter for a new line
Foundation

Tell annsa about your product as you arrive

Onboarding now opens with your product — its name, whether it's a new product, in early use, or running at scale, and a line on what matters most right now. Annsa carries that context into your very first list of priorities. Name and website are inferred from your email domain, so there's nothing to type that annsa can already see. Pick Survey as your first source and you land on the template picker, ready to choose what to ask.

Discovery

NPS, now as a 30-day trend

The NPS card carries a 30-day trend line beside the score and the distribution — each point a 7-day average, so a low-volume survey reads as a clear direction instead of noise. A "now" reading sits next to the line. Until there's enough rated feedback to draw it, the card tells you so rather than showing an empty chart.

Discovery

Read feedback in the words it arrived in

Feedback that came in another language now shows a "Translated from {language} — show original" toggle under the quote. Click it to read exactly what the customer wrote, then flip back to the English the pipeline scores. Across sixteen languages, on every feedback card in a spec or a customer record.

Also this weekControls on tablet now clear a 44px tap target, and the mobile settings nav routes straight to Share Back.

Ask anything, in a space built to keep asking. Review the roadmap in any cadence.

Ask annsa the questions you keep coming back to, and pick up the thread later — every answer opens the real spec or customer beside the chat. Radar now reports daily, weekly, monthly, or quarterly from one toggle, so the same digest powers a Monday standup and a quarterly business review.

Intelligence

Ask — a full canvas for the questions you keep asking

Ask annsa the questions you keep returning to — "what did we deprioritise last month and why?" — and the thread persists, so you can pick it up later. Every answer opens the real spec or customer record beside the chat, not a summary of it.

AskNew conversation
What did we deprioritise last month, and why?
Three priorities dropped out of Next in April. The biggest was Version history — dismissed on Apr 4 after two enterprise accounts churned and the evidence score fell below threshold. The full reasoning opens beside this chat.
Has it come back since?
Yes — 5 new signals this week, across 2 accounts. Radar flagged it under Coming back on Monday.
Ask anything…
Enter to send · Shift+Enter for a new line
Intelligence

Radar reports in the cadence you review in

One toggle switches Radar between daily, weekly, monthly, and quarterly windows. Rising, shipped, and quiet stay the same shape — only the window changes — so the digest that runs your Monday standup also runs your quarterly review.

Discovery

See what to build next for the lens you're working in

A filter row above the kanban scopes the roadmap to a focus engine and a project, and the sequence re-scores to match — so the plan fits the lens you're working in.

Intelligence

Ask annsa answers from the spec or customer in front of you

Open a spec or a customer record and Ask annsa grounds itself in it. Ask "what's the strongest evidence here?" and the answer reasons from that exact spec or customer, not the whole account.

Also this weekAsk annsa threads can be auto-titled, archived, and deleted from a row menu; NPS surveys now record the full 0–10 score natively.

Hear the job to be done behind every customer call. Build the outcome they were reaching for.

Every customer call now tells you the job to be done behind it — the functional job, the outcome wanted, the workaround in place. That job to be done travels into the spec, so a spec explains not just what to build but the outcome the customer was reaching for — and you can see exactly what annsa drew from each call.

Discoverymajor

Transcripts capture the job to be done behind the call

Every transcript is read on two levels — the interview context (who the customer is, the world they work in) and the individual signals inside it. Annsa pulls the job to be done out of the conversation: the functional job, the desired outcome, the current workaround, and where it falls short. That context lands in the customer profile and feeds straight into priority scoring.

Transcript

Discovery call — how the team handles feedback

3 signalsDiscovery
BenProduct leadHelio26 Feb 2026
Context
Who they target
Small product teams shipping every week
Workflows
  • Feedback triage
  • Sprint planning
  • Release notes
Problem space

Turn scattered feedback into a plan the team trusts

Wants: the plan to stay current as new feedback lands

Today: a shared spreadsheet, reviewed on a Friday call

Signals (3)
ChallengeFeedback is scattered across five tools
InsightTriage lives in one person's head
IdeaWants the plan to update as feedback lands
Delivery

Specs explain the outcome the customer was reaching for

With the job to be done in hand, specs frame "why it matters", customer voice, and "done when" around the outcome the customer wanted — not just the feature they named. Quote selection leans toward the lines that prove the job to be done, so a spec reads as a problem worth solving.

Discovery

Trust what annsa pulled from a call

When a transcript finishes processing, you see exactly what came out of it — signals extracted, low-confidence signals skipped, the fidelity rate, new discoveries, and which signals reinforce a priority you already have. Nothing about the extraction is hidden.

Discovery

Open the full transcript from any quote

A quote sourced from a call is now a link. Click "via Transcript" in a spec or a customer record and the full transcript opens in a side sheet with the cited line highlighted in place — the quote in its original context, one click away.

Also this weekRe-uploading the same transcript shows a duplicate notice instead of importing it twice.

Navigate by your products. First priorities ranked for your goal.

Run each product from its own row in the sidebar, with its full workflow underneath. Onboarding learns your goal and your product on the way in, so your first list of priorities is ranked for what you came to do.

Foundationbeta

Run each product from its own home in the sidebar

Each product gets its own row in the sidebar — and you define what a product is: a product line, a SKU, or a project, whatever your team works in. Click the name to land on its priorities, expand it for the full workflow — Priorities, Roadmap, Specs, Customers, Surveys, Share Back. Navigate by the units that matter to you, not by hunting through a switcher.

Foundation

Tell annsa your goal, get priorities ranked for it

Onboarding opens with a north-star focus picker — User growth, Revenue growth, Improve retention, or New features — so your goal shapes the very first list of priorities. Your product name and website are inferred from your email domain, so there's no form to fill.

Also this weekthe Surveys empty state shows five format cards — each opens the wizard with that format pre-selected — and promotes NPS to a hero tile so a new account has a clear first move; saved cuts pin a filtered slice of priorities to the sidebar and Cmd+K.

Specs read like memos. Every spec travels in a link.

The spec panel gets a quiet redesign — rank metadata stops shouting, section dividers come down, and a spec reads as one coherent memo. Share links ship in three tiers, so a spec travels to a teammate, a customer, or the public with exactly the right access. And the Specs page shows you which parts of the codebase the next sprint will land in.

Deliverymajor

See the why behind every spec

Every spec shows the full context behind it — why it ranked where it did, the customer voice that drove it, the reasoning one click away — so you understand the why, not just the what to build. The panel reads like a memo, not a form: rank and metadata sit in a quiet eyebrow, section dividers come down, the spec reads as one document. On a phone it takes the whole screen, so reading a spec on mobile is just reading a spec.

#2in Bug Fixes

Resolve co-edit conflicts in AI rewrites

14 signals·6 accounts·Bug
What to Build

When two people edit a block while AI is rewriting it, merge their changes instead of dropping one. Lock the block to the rewrite, then replay queued edits when it lands.

Customer Voice
"the AI rewrite wiped my teammate's edit while we were both in the same block"
Files to Touch
  • •src/canvas/co-edit.ts
  • •src/ai/rewrite-merge.ts
  • •tests/co-edit.test.ts
Delivery

Share a spec with your team, a customer, or the public

Sharing a spec is one link. Workspace links open for anyone on your team. Company-email links open only for the domains you name, so a customer's team can read a spec without it going public. Public links let a write-up reference a real spec.

Delivery

Specs page — see where the next sprint lands

See at a glance which parts of your codebase the next sprint will touch — a folder-tree heatmap maps the files every open spec will land in. Picking a spec updates the URL, so a Specs link pasted in Slack opens on the same spec.

Discovery

Roadmap carries the work through ship and share

The roadmap kanban now runs past "Next" — Building, Shipped, and Shared columns make the whole post-priority lifecycle visible on one screen. A Start Building action promotes a priority into work as a designed step. Cards use the same two-row shape as the priorities table, so the eye reads one pattern across surfaces.

Also this weekAlerts get their own Settings tab and fire only when a priority graduates into urgent; Share Back deep-links every build into the URL; customer drilldowns survive a refresh and travel as shareable links.

Annsa fits every screen. Every view lives in a link.

Annsa feels right on a phone, a tablet, and a desktop — every screen gets the layout that suits it. Every detail view now travels in a URL, and a priority's spread badge counts companies, not raw feedback.

Foundation

Annsa feels right on any screen

Every page adapts to the device you're on. Phones get a layout built for one-handed reach, with the main destinations a tap away. Tablets get a compact icon rail. Desktop keeps the sidebar, now grouped so settings are scannable. List-and-detail pages collapse to a single pane on smaller screens — reading a spec on a phone is reading a spec, full width. Wide monitors get content caps so lines don't stretch edge to edge.

Foundation

Every detail view travels in a link

Specs, Share Back, and Customers all carry their selected item in the URL. Refresh, the back button, and a pasted link all land a teammate in the same place you were.

Discovery

Account spread counts companies, not feedback items

A priority's account-spread badge now counts the distinct customer companies behind an issue, not the raw number of feedback items. "8 accounts" means eight separate companies are asking — a truer measure of reach than volume alone.

Also this weekSide sheets fill the viewport on mobile and slide up from the bottom; settings tabs cross-fade calmly between sections; the Customers surface leaves beta.

Annsa comes to you. One digest, five places.

Radar lands — a weekly digest that reaches you in the five places you already work. The bell, the email, the Slack message, the Radar tab, and your coding tool all speak one vocabulary: rising, shipped, quiet. Instinct becomes visible — what annsa has learned about how you ship is now a page you can read and adjust.

Intelligencemajor

Radar — your week in review, in five places

Radar gives you one answer to "what happened this week?" Three buckets — rising, shipped, quiet — converge across five surfaces: the Radar tab in the app, the bell notification, the Monday email, the Slack message, and the MCP tool in your editor. Rising items carry a trend badge. Every opted-in account gets the digest at 9am UTC Monday, and a Send to Inbox button delivers it on demand.

RadarWeek of May 12 · Last 7 days
Rising10
Co-edit conflicts in AI rewrites+120%

When two people edit a block while AI rewrites it, changes collide. Five accounts have escalated.

  • @mention notifications+45%

    8 signals · 4 accounts · 2 enterprise

  • Presence on large canvases+22%

    7 signals · 3 accounts

Shipped3
AI thread summary

Shipped 2 days ago

8 of 11 notified · 5 saw it

Inline comment resolve

Shipped 6 days ago

3 of 3 notified

Co-edit conflict fix

Shipped today

Auto-share queued · 4 teams in queue

Coming back2
Workspace permissions
  1. ShippedMar 18
  2. Back nowthis week

8 new signals · 4 accounts · velocity rising

Version history
  1. DismissedApr 4
  2. Back nowthis week

5 new signals · 2 accounts

Intelligencemajor

See and tune what annsa learns from your edits

A page in Settings shows how your decisions have shaped annsa — the narrative ("you've shipped 12 items, mostly bugs, focused on enterprise customers"), your category mix, segment affinity, and domain vocabulary. Controls let you tune how strongly your edits shape future specs, or turn learning off entirely.

Intelligence

Every edit you make teaches annsa

Edit a spec and the save confirmation reads "Saved — annsa learned from this edit." Every PM override teaches annsa how you think, and now the moment the learning happens is visible.

Discoverymajor

Customers carry their full history

Every customer record now surfaces a full product history — every priority their feedback shaped and every spec that shipped to address them. Their personal signal count shows whether "3 mentions" means blocked or just noticed. Annsa treats customers as persistent entities whose voice accumulates meaning over time.

Customers3
Name
Email
Company
Revenue
Last seen
SSimon
Northwind
Enterprise
2d ago
BBen
Helio
Mid-market
1w ago
CCatherine
Meridian
Starter
3w ago

Also this weekAsk annsa weighs your shipping history into its answers, so the recommendation reflects how you ship; every spec archive and stale priority becomes training signal for what you systematically deprioritise.

Priorities show their momentum. The roadmap sequences itself.

A priority is no longer a static row. Every priority now carries state, momentum, and an evidence score — emerging, accelerating, sustained, declining — so the list tells a story. The roadmap turns that into a live sequencing engine that puts the highest-impact, best-evidenced spec at the top.

Discoverymajor

Priorities that show state and momentum

Every priority now has a life of its own. State tracks where it stands — emerging, active, accelerating, sustained, declining, or stale. Signal velocity tracks how fast it's growing. An evidence score combines volume and company spread, so a low-evidence issue stays in "emerging" until the data earns a spec. A nightly sweep keeps every state honest against current velocity.

Priorities3Ready 58Building 2Shipped 12
1
AI rewrite collides when two people edit a block
18 signals · 5 accountsBugRising
2
Presence cursors freeze on large canvases
12 signals · 4 accountsBug
3
Let AI summarise a long thread into a shared docNEW
9 signals · 3 accountsFeature
Discovery

A roadmap that sequences itself by impact

The roadmap is a live scoring engine. Every active priority is scored on effort, confidence, evidence, and fit with your shipping history, then sequenced across a Backlog, Later and Next board. The best-evidenced, best-fitting, quickest-to-ship spec rises to Next. Override any slot — annsa learns from what you disagreed with.

Discovery

See why a priority ranked where it did

Every priority row opens to a score breakdown — urgency, revenue impact, sentiment, competitive mentions — on a clear five-point scale, with the main signal highlighted. The reasoning travels into exported specs under "Why Ranked #N", so the why follows the spec into Cursor, Claude or wherever the builder takes it.

Also this weekReturning users see their priority list instantly with no loading flash; correcting a priority's theme or category cascades a re-classify and spec regeneration automatically.

Read a spec while it writes. Hear Slack the moment it's said.

The first words of a spec appear in about two seconds, with customer quotes on screen while the spec writes itself. Slack listening moves to real time, so a message becomes signal within a second. Specs only generate when the evidence supports them.

Deliverymajor

Streaming specs — first words in two seconds

Specs now stream into the panel word by word. First words appear in about two seconds. The moment you open a priority, the top customer quotes load beneath the skeleton, so you read the customer's own words while the spec writes itself. A time estimate in the activity feed shows how long generation has left.

Discovery

Slack listening, in real time

Slack messages now arrive the moment they're posted instead of on a polling cycle — under a second from message to signal. A thread where three customers discuss the same bug becomes one combined feedback item. Real-time listening is on by default for every connected workspace, no action needed.

Delivery

Specs generate when the evidence supports them

Spec generation now waits for confidence. Emerging priorities and low-evidence issues hold until the data warrants a spec. The moment a priority crosses into solid evidence, its spec generates right away. Every spec carries its trajectory context, so the model knows why this priority matters now.

Also this weekA confidence badge shows spec quality at a glance; a stale badge flags specs that are out of date before you open them; specs regenerate automatically as feedback volume grows or the issue drifts.

A CSV from your tools imports in one step. Nothing submitted is ever lost.

Annsa recognises eleven common tool exports on sight — an Intercom or HubSpot CSV imports in one step instead of three. Reach your plan limit and feedback keeps arriving, held safely and released automatically.

Discovery

A CSV from Intercom, Zendesk or HubSpot imports in one step

Upload a CSV exported from a common tool and annsa recognises it on sight. It knows eleven common export formats — Intercom, HubSpot, Zendesk, Typeform, Freshdesk, Canny and more — so when the headers match, the import wizard skips column mapping and lands straight on preview: "Intercom export detected."

Foundation

Every signal is kept, even past your plan limit

Reach your monthly limit and feedback keeps arriving — every item is held safely in a queue and released automatically the moment your plan renews, tops up, or upgrades, oldest first. The Billing tab shows exactly what's waiting and when it lands, so you always have the full picture.

Also this weekFeedback that arrives with a name but no email still creates and links a customer record across every channel; connected Google Sheets auto-import new rows with no Sync click.

Onboarding that fits your role. Settings pages you can share.

A short, calm week. Onboarding asks who you are on screen one and tailors the rest. Every settings section becomes its own URL — share a Billing link, bookmark Team, refresh without losing your place.

Foundation

Role-aware onboarding tailors your start

Tell annsa your role and the rest of onboarding shapes itself to match. GitHub benefits show inline before you connect, so the value is clear by the time you reach for the button.

Foundation

Settings pages get their own URLs

Each settings section becomes its own page. Deep-link to Billing, use the back button, refresh without bouncing out. Centred on wide screens.

Also this weekPriority groupings sharpen at larger volumes, batch park and restore stay applied, mobile priorities scroll smoothly, the activity log moves to its own page, and Get Feedback is reachable from the avatar dropdown anywhere.

Every source in one menu. Surveys in five steps.

Connect any feedback source from one consolidated menu, with connection status visible at a glance. Surveys ship with a five-step setup wizard. Briefs hold their edits on every refresh.

Discovery

+Add brings every source into one menu

Six feedback sources, one click to connect each. Connection ticks show what's live without going into Settings. Clicking Slack opens the channel wizard directly from anywhere.

Discovery

Survey setup in five guided steps

A five-step wizard takes you from "what type" to "embed code" in minutes. Drafts auto-save so you can finish later. Response counts update live as feedback arrives.

Delivery

Briefs hold their edits on refresh

Manual edits stay where you put them — only untouched sections regenerate. GitHub context caches per run, so generation runs faster.

Discovery

Priorities that catch subtle differences

Priority groupings read nuance that similar phrasing used to merge. "Slow search for admins" and "slow search on mobile" stay separate when they should.

Every customer, 16 languages.

Feedback in any of sixteen languages now lands in one ranked list. Every customer who has written in gets a record, with revenue context feeding straight into scoring. Ask annsa answers from all of it — and a free plan opens the door from signup.

Discoverybeta

Sixteen languages, one priority list

Feedback in any of sixteen languages translates into English for analysis and contributes to the same ranked list. The customer's original quote stays preserved in the brief, so engineers read the customer's exact words. Supported alongside English: Spanish, French, German, Portuguese, Japanese, Korean, Chinese, Italian, Dutch, Arabic, Hindi, Russian, Turkish, Polish and Swedish.

Discoverybeta

Customers — every submitter, with revenue context

Every customer who has sent feedback gets a record. Filter by revenue band. See everything a customer has submitted. Their voice feeds directly into priority scoring — the foundation for treating customers as persistent entities whose voice accumulates meaning over time.

Customers3
Name
Email
Company
Revenue
Last seen
SSimon
Northwind
Enterprise
2d ago
BBen
Helio
Mid-market
1w ago
CCatherine
Meridian
Starter
3w ago
Intelligence

Ask annsa — answers from your data

Ask in plain language — "which priorities have the most enterprise backing?" — and annsa answers from your product history. Thumbs up or down trains future answers.

Foundation

Free plan from signup

Pick the free plan upfront, no credit card. Plan choice happens before you enter the app.

Also this weekPark any priority from the menu or a keyboard shortcut, with batch park supported. The activity feed badge shows new items since your last visit.

One app. Many streams.

A persistent sidebar makes annsa feel like one app on every page. Projects give each product line its own feedback stream. Surveys ship as the umbrella for five feedback formats.

Foundation

Persistent sidebar across every page

Navigation always visible. Ask annsa, breadcrumb headers, and the avatar dropdown one click away. Nav icons match the marketing site, so product and website feel like one brand.

Discoverybeta

Projects keep each stream's priorities focused

Each product line, team, or initiative gets its own feedback container. Slack channels map to projects so signal routes correctly from ingestion. Clustering respects project boundaries — a billing request never competes with a dashboard feature.

Discovery

Surveys — five formats, one umbrella

Five survey types — Bubble, Embed, Banner, Thumbs, Trigger — each suited to a different moment. Per-survey analytics show sentiment, revenue impact, and which pages generate the most signal.

Slack across every channel. Surveys that capture the why.

Slack listens across every channel you pick. Surveys capture written feedback alongside every reaction. Priority titles describe consequences instead of topics. Clusters tighten around your team's vocabulary. MCP gains a circuit.help command.

Discovery

Slack listens across every channel you pick

Connect Slack and pick the channels that matter. Messages arrive deduplicated, so the same point never gets counted twice. The activity feed updates as feedback flows in and specs flow out.

Delivery

MCP — circuit.help in your editor

Ask Cursor or Claude Code what annsa can do. circuit.help lists the tools and orients the first connection.

Discovery

Surveys capture the why behind every reaction

Every survey type now captures written feedback alongside the reaction — the reason underneath the rating. All surveys follow your customer's system theme.

Discovery

Priorities titled by the consequence

Priority titles read like consequences — "Slow search frustrates power users" — not topics. Four intent types keep clustering tight.

Intelligence

Clustering learns your vocabulary

Annsa picks up your product's terminology from your website and feedback corpus. Intent-anchored embeddings produce tighter clusters. Parameters auto-tune per account.

Also this weekFaster mobile load times, OAuth and error handling sharpened across GitHub, Slack, and Sheets, and manual feedback now captures customer name, email, and revenue band.

Priorities surface as they're scored. Briefs generate faster.

The pipeline runs in the open with a live activity feed — priorities appear before specs finish. Brief generation gets faster while keeping the quality bar. Focus lenses explain what each one optimises for.

Intelligence

Pipeline activity — see priorities as they generate

The pipeline runs visibly. Priorities appear before briefs finish. Per-brief progress reports 1/5 → 2/5 → 3/5. The activity log paginates, so you can browse older activity without leaving the page.

Delivery

Briefs generate faster

Spec generation finishes faster while keeping quality. Refresh a brief when new feedback arrives.

Discovery

Focus lenses explain themselves

Each focus lens names what it optimises for, so the team can see why retention mode ranks differently from growth mode and switch with confidence. Assignee surfaces in the priority table and brief panel.

Foundation

Calmer dark mode, native Sheets picker, Stripe and Google Pay

Dark mode transitions smooth out. Google Sheets opens the native Picker after OAuth. MCP setup goes zero-config. Stripe and Google Pay land in iframe checkout.

Also this weekSeparate staging and production environments, and cookie preferences saved permanently across visits.

Customer calls become priorities. Shipped becomes shared.

Transcripts ship — one call becomes up to twenty-five classified feedback items, the customer's actual words preserved into the brief. Surveys ship as the beta umbrella for five feedback formats. Share Back completes the loop.

Discoverybetamajor

Transcripts turn customer calls into priorities

Upload a recording or paste a transcript and annsa pulls out the feedback. Speaker attribution, topic segmentation, churn signals, competitive mentions. One transcript becomes up to twenty-five classified items, the customer's actual words preserved through to the brief.

Transcript

Discovery call — how the team handles feedback

3 signalsDiscovery
BenProduct leadHelio26 Feb 2026
Context
Who they target
Small product teams shipping every week
Workflows
  • Feedback triage
  • Sprint planning
  • Release notes
Problem space

Turn scattered feedback into a plan the team trusts

Wants: the plan to stay current as new feedback lands

Today: a shared spreadsheet, reviewed on a Friday call

Signals (3)
ChallengeFeedback is scattered across five tools
InsightTriage lives in one person's head
IdeaWants the plan to update as feedback lands
Discoverybeta

Surveys — five types for five moments

Five survey types — Bubble, Embed, Banner, Thumbs, Trigger — each suited to a different moment in the product. A slide-out editor with live desktop and mobile preview. Per-survey analytics show sentiment, revenue impact, and top page locations.

Delivery

Share Back closes the loop in one sheet

Tell every customer who asked for a feature when it ships. Approve, skip, or customise per recipient. Choose email, a banner, or both. A new "Shared" status marks the full circuit complete.

Foundation

Settings — four sections, one page

Sign in with GitHub alongside Google. Settings consolidated into four sections — Account, Integrations, Team, Billing. Promote an editor to owner from the team roster.

Also this weekA 14-day free trial on every plan with self-service upgrade, downgrade, and cancellation, plus CSV uploads and GitHub OAuth sharpened.

Annsa Log opens up. Search by meaning. 2FA at the door.

Annsa Log makes pipeline processing visible — every theme, every intent breakdown, every brief generated, in real time. Semantic search lands across the app from Cmd+K. Two-factor authentication ships before the first user signs in.

Intelligence

Annsa Log — watch the pipeline as it runs

A real-time view of everything annsa does — the current theme, intent breakdown, elapsed time, every spec generated as it generates. The activity timeline retains ninety days.

Activity Log5
Generating specs
Today
114:31:58+12 pieces of feedback imported: Slack, CSV upload
214:32:13›Classified by intent: 8 features, 3 bugs, 1 praise
314:32:21◇Pattern forming: Onboarding drop-off
414:32:41Generating spec
514:32:49✓Spec ready: Faster first-run setupView spec →
Foundation

Two-factor at the front door

Two-factor authentication with recovery codes lands before the first user signs in. A pre-launch security review covers authentication, the service layer, and content security.

Discovery

Score the feedback already in your Google Sheet

Import from Google Sheets — annsa maps the columns automatically, and the feedback flows straight into scoring and ranking.

Foundation

Search by meaning, Cmd+K from anywhere

Search treats "slow" and "performance" and "loading time" as the same concept. One search finds every variation. Cmd+K opens the command palette from any screen.

Also this weekExport briefs as Markdown or full data as JSON, mobile OAuth sign-in, priority summaries in problem-statement format, and duplicate priorities that catch and merge after save.

Memory ships. Annsa learns from every brief.

Memory ships — every brief you ship teaches annsa how you think. Theme, customer segment, and category are recorded as memories that shape what surfaces next. The Priority Engine sharpens with website-aware classification, and multi-select makes batch actions painless.

Intelligencemajor

Memory — annsa learns from every brief you ship

Annsa now writes a memory every time you ship. Theme, customer segment, category — recorded and ready to shape what surfaces next. Priorities that match your shipping history get flagged. Briefs surface what annsa remembers — past ships, corrections, related context. The longer annsa runs, the sharper its sense of your product becomes.

Discovery

Smarter priorities, in your team's language

Connect your website and annsa classifies feedback in your product's terms. Correct a category and it sticks across new feedback. Competitor mentions appear on the priority card.

Discovery

Batch actions — one action across many priorities

Select multiple priorities and act on the whole set. Change status, assign, or export as Markdown. Undo a CSV import from the completion screen.

Foundation

First priorities in minutes

Pick your sources, connect GitHub, and the first priorities arrive in minutes. A live processing banner shows stage-by-stage progress.

Launch week. Feedback in. Specs out. The loop closes.

The first version of annsa. Priorities ranked from feedback across six dimensions. Specs that reference your codebase, ready for an AI coding agent before standup. MCP pulls priorities and specs into Cursor and Claude Code. When a feature ships, every customer who asked for it hears back.

Discoverymajor

Launch — score what customers are asking for

Connect your sources. Priorities scored across six dimensions — volume, urgency, revenue impact, positive sentiment, negative sentiment, feature demand — ranked and ready in minutes. A weekly digest tells you what moved, what's new, and what dropped off.

Priorities3Ready 58Building 2Shipped 12
1
AI rewrite collides when two people edit a block
18 signals · 5 accountsBugRising
2
Presence cursors freeze on large canvases
12 signals · 4 accountsBug
3
Let AI summarise a long thread into a shared docNEW
9 signals · 3 accountsFeature
Discovery

Surveys — catch feedback inside your product

A one-line embed lets customers share feedback inside your product. Feedback flows straight into annsa for scoring and ranking. The theme matches your brand.

Discoverymajor

Priority Engine — pick the goal that ranks your list

Set a goal — Revenue Growth, Retention, Bug Fixes — and every priority re-ranks to match. Switch lenses anytime. The list reflects the goal you're optimising for.

Deliverymajor

Codebase-aware specs — written in your conventions

Connect GitHub and every spec knows your codebase — file structure, tech stack, naming conventions, recent commits. The result references real file paths, ready for Cursor or Claude Code via MCP. Annsa becomes the first product tool whose primary user is a coding agent.

#2in Bug Fixes

Resolve co-edit conflicts in AI rewrites

14 signals·6 accounts·Bug
What to Build

When two people edit a block while AI is rewriting it, merge their changes instead of dropping one. Lock the block to the rewrite, then replay queued edits when it lands.

Customer Voice
"the AI rewrite wiped my teammate's edit while we were both in the same block"
Files to Touch
  • •src/canvas/co-edit.ts
  • •src/ai/rewrite-merge.ts
  • •tests/co-edit.test.ts
Delivery

MCP for Cursor and Claude Code — specs land in your editor

Pull priorities and specs directly into Cursor or Claude Code. The spec is there when the editor opens. The format is built for AI coding agents.

Deliverymajor

Close the loop — every customer who asked, told

Mark a brief as shipped and annsa emails every customer who asked for it, with their original feedback quoted back. Priority status tracks ready, building, and shipped.

Also this weekBrief versioning with revert to any version, PII stripped before every AI call, dark mode following system preference, and team invites with owner and editor roles.

Ship. Hear back. Ship again.

Try Annsa

New here? See how the circuit works →