Selected work / Enterprise AI
Engineering MeshMesh
through to acquisition.
Form From helped build and harden MeshMesh’s AI platform for Salesforce. Our contribution spanned platform engineering, security and technical due diligence during the company’s acquisition journey.
- The product
- An AI agent platform that carries out Salesforce tasks from natural-language instructions.
- Our contribution
- Platform engineering, security remediation, evaluation and technical due-diligence tooling.
- The milestone
- MeshMesh was acquired by Salesforce. Form From contributed to the engineering and diligence work.
The task
Make an agent dependable inside a business system.
MeshMesh lets people describe Salesforce work in plain language and have an agent carry it out. That makes the result in the application.not just the quality of a response.the thing to assess.
We worked within the engineering team as the platform developed and underwent technical due diligence. The work involved both improving the software and helping reviewers understand its behaviour.
Check what changed.
- 01
A task and an expected result.
- 02
The agent works in the target system.
- 03
Check the resulting system state.
- 04
Record the result and investigate failures.
Our contribution
Build, test and make the result inspectable.
Form From contributed to platform security, authentication behaviour, reliability and evaluation. We also built automated technical due-diligence tooling and prepared capacity evidence for review.
The evaluation approach ran tasks against Salesforce environments and checked the resulting system state. This made it possible to examine whether an agent had completed the requested work and investigate where it failed.
Security remediation and technical reporting accompanied the build. We documented measured behaviour and distinguished it from limits that had not been tested.
The result
A contribution through to acquisition.
MeshMesh was acquired by Salesforce, with Form From directly involved in the engineering and technical diligence work. The acquisition was a company milestone achieved by the wider team; our part was to help build, strengthen and document the platform.
What this means for another project
When AI can change a system, test what changed.
We assess the action where it happens: the record created, the task completed or the state updated. That principle applies wherever AI is given permission to act inside business software.
Discuss a similar project