Case studies
Proof, not promises.
A closer look at three systems we built, one keeping safety-critical robotics audit-ready, one bridging telephony to the CRMs businesses already run, and one turning a platform API into an AI concierge over the Model Context Protocol.
Case study 01 · Roboticks
Audit-grade verification for ROS2 robotics.
Roboticks is a continuous-integration platform purpose-built for robotics software verification and validation, turning safety compliance from a spreadsheet chore into an automated, provable part of the pipeline.
- ROS2
- GitHub Check Runs
- Gazebo Harmonic
- Webots
- ReqIF
- Python & C++ SDKs
The challenge
Teams building safety-critical robots, cobots, AMRs, ADAS, surgical systems, eVTOL, have to prove their software meets strict standards. That verification traditionally lives in spreadsheets that drift out of date the moment the code changes, and audits become a scramble.
What we built
A CI platform in three parts: automated test execution on every pull request via GitHub Check Runs; requirements traceability that maps safety requirements to tests with @confirms decorators and ReqIF round-trips into Polarion, Jama and DOORS; and hash-chained Evidence Packs bundling JUnit, SARIF, ReqIF and MCAP data.
Hosted GPU simulation (Gazebo Harmonic, Webots), self-hosted compute, and Python and C++ SDKs round out the platform.
The outcome
Traceability that can’t go stale, offline-verifiable evidence ready for regulators, and coverage across 29 tier-1 safety standards, no spreadsheets required.
Case study 02 · Bicom Systems
Connecting telephony to the CRMs businesses already run.
Bicom Systems, a global Unified Communications and VoIP provider, partnered with SigmaBit to bridge its telephony platform to the CRM systems its clients already run.
The challenge
Clients increasingly manage customer interactions in CRM software. Bicom wanted seamless integration between telephony and those platforms, but Odoo, a modular ERP, and BlueCRM, a specialized customer-management tool, each needed a different strategy.
What we built
A flexible, API-driven middleware layer that moves both structured and unstructured data reliably between telephony systems and CRM backends, engineered for scalability, error handling and monitoring across distributed environments.
The outcome
Out-of-the-box CRM connectivity that shortened onboarding, reached new markets, and laid the foundation for AI-powered features: real-time transcription, sentiment analysis and predictive lead scoring.
Case study 03 · Picklebot
An MCP server that answers plain-English questions from live data.
Picklebot is an AI concierge for the Pickleball.com ecosystem: an MCP server in front of the platform API that picks the right tool, queries live data and streams back an answer in natural language.
- Model Context Protocol
- AI agents
- Tool calling
- Streaming SSE
- Live platform data
- Read-only by design
The challenge
Everything a player wants to know, tournaments open for registration, clubs nearby, team and ladder leagues, player ratings, already exists in the platform API, but only if you know which filters to set. People ask in sentences; the API answers parameters.
What we built
An MCP server exposing six read-only tools, tournaments, players, clubs, team leagues, ladder leagues and a support FAQ. A question is routed to the right tool, its parameters extracted from the sentence (rating, format, gender, location, status), live data queried, and the answer streamed back as natural language.
A second mode returns the same result as structured JSON, so product surfaces can render cards instead of prose. Scoped authentication with cached tokens, location awareness and read-only access are built in.
The outcome
One server that answers a person in chat and an application over an API, and a reusable pattern for making any platform AI-addressable. Picklebot is in development and being integrated into the Pickleball.com experience.