Opportunities › Audit trail and compliance evidence for AI agents › Proposed app
Tallyseal
Signed, independently verifiable records of what third-party AI agents did
- Score
- 33
- confidence 49
- Evidence
- 6
- items on the page
- Rivals
- 0
- named in sources
- First version
- 45 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2024
- New market · AI estimate
ScreensConcept
Desktop app: Compliance teams and auditors review detailed run records, evidence bundles, and hash chains to verify agent behavior—back-office work requiring focused desk time. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
Live run
Sealed142 actions
About this appAI proposal
Tallyseal records every tool call, model call and data access made by the AI agents your vendors run for you, and seals each record with a signature and a hash chain. Your auditor can check the records without trusting the vendor's own logs.
It is built for companies that deploy third-party agents and must show SOC 2 or ISO 27001 auditors what those agents really did. Records capture what happened on each run, so you do not have to assume the same prompt behaves the same way twice.
Instead of a vendor dashboard you cannot verify, you get an evidence bundle: sealed run records, a verifier an auditor can run alone, and a plain report mapped to the controls you are asked about.
- Tagline
- Proof of what your AI agents did, checkable by anyone.
- Voice
- PreciseCalmPlain
- Why this name
- A tally counts every action and a seal marks it as untouched, like a wax seal on a record. Two plain words joined.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Tamper-evident run records
Each action is hashed and chained to the one before it. Any edit, gap or reordering shows up when the chain is checked.
No member yet
Independent verifier
Auditors check a bundle with a small open verifier and your public key. No vendor login or vendor-supplied log is needed.
No member yet
Gateway capture
Route agent traffic through a proxy that logs model calls, tool calls and data access outside the vendor's control.
No member yet
Per-run behavior view
Compare runs of the same task side by side to see where actions differed, rather than relying on one past log.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#12372FInk Green · primary#F6F3EAPaper · surface#1D2421Charcoal · text#D98B1FWax Amber · accent#7FA88FSage · data
Typography
Aa
Tallyseal
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Proof of what your AI agents did, checkable by anyone.
PricingAI proposal
The material gives no competitor prices. These tiers, with per-agent licensing from the monetization idea, are a starting point to test.
Starter
Free
- 1 agent
- 7 days of sealed records
- Verifier download
Team
$199 a month
- Up to 10 agents
- 12 months of records
- Run compare and data map
- Evidence bundle export
Enterprise
Contact us
- Unlimited agents
- Custom retention
- Own signing keys
- Integration support
Would you pay Team · $199 a month?
Your answer tells PWAS which prices people accept. It changes no score.
No answers yet.
What it could be worthEstimate
What a software business making this much sells for today, from real deals. Pick a monthly revenue.
If it makes
Worth about
$194,000
Sale range $132,000 to $263,000
$120,000 a year in revenue
From 87 software businesses listed and sold on Empire Flippers in the last year, revenue vetted by the broker: they sell for about 19× their monthly revenue, most between 13× and 26×.
Measured on profit, the middle deal is 32× monthly net profit: about 2.7 years of profit.
An estimate, not an offer or financial advice. A real price depends on profit, growth, churn, how much of the work is the owner's, and the buyer.
How it would growAI proposal
How the product would reach new customers through its own use, and how the first ones would be found. A plan to test, not a result.
- Built-in loop
- Shared outputs. Customers send evidence bundles to auditors and to vendors' security reviewers. Each bundle carries the verifier and a Tallyseal-sealed mark, so recipients see how it was made.
- Do they become customers?
- Auditors and vendors who open bundles could become buyers, but this is unproven. Test whether auditors recommend it to other clients, and whether vendors ask for it in security reviews.
- First customers
- Security and compliance leads at 50 to 500 person software companies pursuing SOC 2 that already run one or more third-party AI agents with access to customer data.
- Where to reach them
- Hacker News and security forums: Share a working verifier and a write-up on why vendor logs are not enough, then invite readers into a pilot.
- SOC 2 auditors and advisors: Ask small audit firms to review a sample bundle and say what they would need to accept it.
- Compliance automation communities: Offer a free Starter setup to teams asked about AI agents in their audits.
- First milestone
- Five paying customers in one segment, plus one auditor who accepts a Tallyseal bundle as evidence in a real audit.
Why people already pay for this
- VerifiedHacker News post received 37 points and 23 comments
- Public claimEnterprises cannot verify what AI agents do with their data when deployed from vendors
- Public claimVendor-supplied audit logs and dashboards are not reliable for compliance
- Public claimTraditional compliance frameworks are inadequate for agentic systems
- Public claim
Tech stack
Every proposal builds on the same PWA base, installable from the browser on any phone or computer. Dashed boxes are what this app adds.
- AppNext.js and TypeScript
- PWAManifest and service worker
- UITailwind CSS
- Data and authSupabase Postgres
- PaymentsStripe
- EmailResend
- HostingVercel
- This app · Capture gatewayLightweight proxy for model and tool trafficRecords must be taken outside the vendor's systems to be independent of their logs.
- This app · Signing and key storageManaged key service with signing supportSigned records need keys the customer controls, with rotation and audit.
RisksAI inference
- TechnicalCapturing agent actions outside the vendor needs a proxy in the traffic path. Some agents run inside vendor clouds and cannot be routed, which limits coverage.
- CommercialAuditors may not yet accept this kind of evidence. It must be tested with real SOC 2 and ISO 27001 auditors before buyers pay.
- DataRecords may contain sensitive prompts and customer data. Storing them creates a new risk and needs redaction and strict access control.
- PlatformLarge agent platforms may add their own signed logs, though the evidence says vendor-supplied logs are seen as less trusted.
- RegulatoryRules for agent evidence are still unsettled, so the report mapping could be out of date or claim more than auditors accept.
Everything marked AI proposal or AI inference was written by a model from the evidence on the opportunity page. The proposed name must be checked against trademark registers before it is used.