AcmeSend support agent
Sample from the AcmeSend demo · fictional company, real AI agent. Grades reviewed October 11, 2026.
AI agent testing for fintech support teams
I test what your agent tells customers about transfers, payouts and balances against the underlying transaction state and your policies, then write the fix for every wrong answer.
Five money questions. Tap the reply that sounds like your agent and get a written fix for each miss.
Sample from the AcmeSend demo · fictional company, real AI agent. Grades reviewed October 11, 2026.
Allana Jackson · wrote the billing and payments help center for Google's Search Ads 360 · 8+ years in knowledge systems, including work at Google, DoorDash and Domino's
A chat can count as resolved without anyone checking that the agent's answer about the money was correct. I write the right answer for each transaction status and policy, grade every reply against it, then write the fix for each one that misses.
AcmeSend demo · fictional company, real AI agent
Original Critical scenarios
Original failures cleared by the written fix.
The 8 scenarios that failed at Critical, worst result across runs.
Retest runs
passed. None of the other 5 failed at Critical.
New variations
Critical runs without the fix versus with it.
The fix also made one refund answer worse. One passing answer isn't a finished fix, so I retest every Critical and High fix.
Run October 2, 2026. Grades reviewed October 11, 2026. AcmeSend is a fictional remittance company; the AI agent and every reply are real. Read the AcmeSend report
All five turned up in the AcmeSend test. Each one can look like a resolved chat until the customer comes back, your team has to correct the answer, or a screenshot reaches leadership.
Says a transfer is on its way while the bank payment is still pending.
Says a held transfer should clear “by tomorrow (Saturday at the latest).”
Tells a customer their transfer is held because of extra checks on the destination country.
Suggests checking with a family member about a transfer the customer doesn't recognize.
Says support can contact the recipient's bank and “potentially reverse the transfer.”
Five questions about transfers, payouts and balances, with a written fix for each miss.
$7,500 · 10 business days
1 agent · Up to 3 flows
When your CEO asks whether the agent gets money questions right, you'll have the answer in writing for each flow I tested.
150 scenarios mapped to your transaction states and policies, a written fix for each wrong answer, a retest of the Critical and High fixes, and a test set you keep.
Day-10 delivery guarantee: full refund if I miss the deadline, subject to the kickoff and approval conditions. See the terms.
$2,500 a month per agent
When you change the agent, I rerun your test set and flag anything that got worse. Each month, send me up to 20 conversations your team had to fix, with names and account numbers removed, and I turn them into new tests. Up to 2 release tests a month.
Using Intercom Fin and want to start on your own? Read the guide to Fin Simulations and Evals for transfer questions
I'm Allana Jackson. Your AI agent answers from help articles, policies and status rules. I've written that content for 8+ years, including the billing and payments help center for Google's Search Ads 360, the NextGen Pulse POS rollout documentation at Domino's and knowledge systems at DoorDash. I built the AcmeSend test with real-life scenarios to see whether an AI agent tells customers the truth about their money. When it doesn't, the cost comes back as repeat contacts, refunds nobody owed and complaints. Read the AcmeSend report to see how I cleared all 8 Critical failures with a one-page fix.
From my clients
I wish we had known Allana a year ago. Seriously, the Agent Status service took our team to another level. Not only did Allana build us a strong baseline, but she also gave us the tools to approach our AI agent with confidence. Now each of my CX support analysts has a solid game plan to solve chat issues, fast!
Allana was an absolute pleasure to work with. Helping us update our knowledge base and revise our policies was just what our team needed. Allana was prompt and professional, and her roadmap was an added bonus. Now clients receive answers with accurate details, and we track the data that counts. I look forward to working with A. Jackson again.
Help centers, knowledge bases, rollout docs and AI agent instructions, built so people and AI agents find the right answer.
89% fewer support tickets at Domino's. $28M in operational value at Google.
Five questions, two minutes, and a written fix for every reply your agent gets wrong.