AP3 at the Accenture TechNext Challenge 2026
AP3 won the Cybersecurity track at Accenture’s TechNext Challenge, bringing private agent collaboration to an enterprise innovation stage.
Written by
Jay Prakash
Bengaluru, India
Aug 18, 2026
AI agents are starting to work across organisational boundaries. To make those workflows useful, enterprises need a way to compute with sensitive information while keeping control of the underlying data.
At the Accenture TechNext Challenge 2026 finals in Bengaluru, Silence Laboratories presented the Agent Privacy-Preserving Protocol, AP3. We announced our selection as the Cybersecurity Track Winner on 18 August.
The enterprise problem behind AP3
A collaborative workflow can involve financial records, credentials or risk signals held by different parties. Granting access to that information creates obligations that continue throughout its use.
AP3 applies secure multi-party computation to this setting. Participating agents can work towards an agreed result while protecting their private inputs. The design combines computation with cryptographic identity and independent approval policies, so that the parties can define the work they permit.
Making privacy usable for builders
The engineering challenge includes helping developers express useful workflows and understand their boundaries. Our approach compiles computation into an auditable circuit, with the release of results governed by the agreed workflow.
For a team designing an agent system, that shifts attention to concrete questions: what should be computed, whose approval is required and which output is necessary for the next action?
The TechNext Challenge gave us an opportunity to present this approach in an enterprise innovation setting. The recognition supports our focus on making private collaboration practical for the teams building the next generation of agent applications.
