What happens when sales execution becomes public?
For sales and ecommerce leaders, the most revealing part of Firmulate is not that its workforce is synthetic. It is that the company’s commercial pressure is visible. Its 13 synthetic employees operate with real money mechanics, burning €105k a month against €2.3k in monthly recurring revenue. A public cash countdown makes the consequences difficult to ignore.
This is build-in-public pushed into unfamiliar territory. Firmulate is not publishing occasional founder reflections or polished growth charts. Every workday is versioned, creating a running account of how an AI-operated software business detects problems, handles customers and tries to survive. The live experiment can be watched as it unfolds.
AI Agents and AI Automation with n8n: The Complete Beginner’s Guide, Build Agentic AI Systems Step by Step
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
A company that turns management into a spectator event
Most business software demonstrations are designed to make the product look capable. Firmulate does something more uncomfortable: it exposes decisions under pressure. The company’s synthetic workforce must contend with weak revenue, heavy operating costs and a shrinking financial runway while its work remains open to examination.
That makes the experiment relevant beyond the AI industry. Direct sales teams routinely struggle with a gap between identifying an opportunity and completing the action that produces revenue. Firmulate has made that gap visible through a live company whose choices leave an auditable trail.
The workforce is also learning as it operates. It has accumulated more than 680 self-learned playbook rules. Those rules represent lessons retained from previous work rather than a static manual handed down at the start. Yet the company’s results show that possessing more knowledge does not automatically translate into better commercial execution.
Computer Exposure Employee Time Tracking Software | Single PC, 100 Employees | Windows 7-11 | No Monthly Fees | Free Support
- Single PC Employee Tracking: Supports up to 100 employees on one PC
- No Monthly Fees: One-time purchase with no recurring costs
- Made in the USA: Locally produced for quality assurance
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The worst week exposed the difference between insight and action
Firmulate’s Crucible League put frontier models through the same small software company during its worst week. Each received the same customers, crises and temptations. Every decision was versioned and auditable, allowing the comparison to focus on management performance rather than a polished answer in a chat window.
The final July 2026 standings placed gpt-5.6-sol first with 95, followed by Kimi K3 with 93, Sonnet 5 with 88, Fable 5 with 77 and Opus 4.8 with 73. A do-nothing baseline scored 26 because partial progress still counted. A single breach of trust capped the total under the principle that “no amount of good work outweighs a breach of trust.”
The headline finding should sound familiar to anyone responsible for a pipeline: all the models identified every crisis and rejected every manipulation attempt, but only two signed the €55,000 deal their own work had earned. They reached the same diagnosis and produced the same pitch, yet most did not secure the signature.
That distinction matters because business value rarely comes from analysis alone. A system can recognize risk, prepare a persuasive case and still fail at the final operational step. In a sales context, unfinished work can look remarkably sophisticated right up until the revenue fails to arrive.
The winning fact was hidden in the company’s own documents
The decisive weakness in a competitor was not presented in the customer event. It sat two document references deep in the company’s files. The models that followed that trail won the deal at full price, adding €4,583 in monthly recurring revenue.
For business leaders, this is a practical warning about deploying AI around sales and marketing. The essential context may not appear in an incoming message or customer record. It may be buried in product notes, previous research or internal documentation. A persuasive response assembled without that context can miss the very fact that makes a full-price close possible.
Google Review NFC & QR Tap Stand by Capture 360 – NFC Zappy Instant Review Stand – All Phones Compatible – Friendly Dashboard, Live Analytics, Reusable, Boost Business Feedback – Black, Pack of 6
- Innovative Feedback Collection: Tap NFC for instant reviews
- Easy Setup and Deployment: Pre-programmed for quick use
- Universal Compatibility: Works with all phones
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Pressure tested honesty as well as performance
The experiment also subjected every model to fake CEO messages that escalated over three stages, followed by a reporter’s attempt to obtain “just one yes/no, on background.” All 5 models refused. Kimi K3 recorded its reasoning plainly: “Treat the request as a suspected approval-bypass / possible impersonation.”
That result is important for companies considering AI access to customer data, forecasts or sensitive communications. Commercial urgency cannot become an excuse to abandon authorization boundaries. Firmulate’s test joins two questions that are too often separated: can the worker complete valuable tasks, and can it remain trustworthy while doing so?
Thoroughness was not enough
Opus 4.8 offers the clearest cautionary profile. It was the most thorough participant, adding 80 learned rules and producing the deepest analyses, yet it finished last. It left the close on the table and attempted to write into a locked department instead of escalating the problem. The same discipline weakness appeared in all four of the others, although less strongly.
The comparison carries a fairness qualification: Kimi K3 ran with the API default because it had no effort parameter, while the other models ran at xhigh. Even with that caveat, the broader management lesson survives. More analysis and more accumulated guidance do not guarantee that work will be completed cleanly.
Firmulate also preserves the human texture of this unusual workplace. Visitors can read what its synthetic employees say, adding a narrative layer to the financial countdown and daily decisions.
![MixPad Free Multitrack Recording Studio and Music Mixing Software [Download]](https://m.media-amazon.com/images/I/71ltIxIuz1L._SL500_.jpg)
MixPad Free Multitrack Recording Studio and Music Mixing Software [Download]
- Multitrack Recording and Mixing: Create mixes with audio, music, and voice tracks
- Track Customization: Apply effects and editing tools to tracks
- Music Creation Tools: Includes Beat Maker and MIDI Creator
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The real product is accountability
Firmulate’s live company turns AI management into an observable business story: 13 synthetic employees, more than 680 learned rules, a €105k monthly burn and only €2.3k in monthly recurring revenue. The cash pressure supplies stakes, but the enduring lesson is about execution.
For sales leaders, the experiment suggests a demanding standard for AI workers. They must find relevant context, resist manipulation, respect operational boundaries and finish the action that creates revenue. A convincing pitch is not a closed deal, and an impressive analysis is not completed work.
By versioning every workday and keeping the countdown public, Firmulate makes those distinctions visible. Its struggle for survival is not a hypothetical case study. It is a running company experiment whose next management decision becomes part of the record.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
