LLMs & AI Agents · Public affairs
When Information Pressure Rises, an AI Agent Creates Order
New legislative proposals, position papers, background documents and stakeholder statements appear every day. For public affairs consultancies that means huge volumes of information, high time pressure and little room for manual routines. RPP Group responded with its own AI agent.
Julia Rose
Key takeaways
- A large share of RPP Group’s working time went into recurring tasks: reviewing documents, summarising, comparing positions, formulating key messages.
- Together with Policy-Insider.AI we built ChatRPP, a system of five specialised agents tailored to how the consultancy works.
- Time spent on manual routine work fell by around 70%, answers arrive in real time and first drafts in seconds.
- Assessment and strategic decisions stay with people, and data never leaves the firm’s own infrastructure.
The starting point: routine crowds out strategy
RPP Group’s daily work showed a familiar pattern. A significant share of working time went into recurring tasks: reviewing documents, summarising content, comparing political positions, formulating key messages. These tasks were necessary but tied up strategic capacity. In public affairs consulting, where context, experience and judgement are decisive, that left less time for client-facing work.
What was needed was a solution that understands how the consultancy works, hits the professional tone, meets compliance requirements and fits into existing workflows. An off-the-shelf tool could not have done that.
The solution: ChatRPP
Together with Policy-Insider.AI we developed ChatRPP, an AI system for public affairs. Agentic AI refers to systems that handle tasks in context, carry out several steps in sequence and deliver structured results. ChatRPP consists of five specialised agents: an orchestrator routes each request to the right specialist, and dedicated QA agents check results against internal standards. The system supports the team in
- analysing policy documents,
- summarising complex content quickly,
- deriving strategic options,
- developing clear key messages
- and producing communication material in the consultancy’s style.
Users can ask follow-up questions, upload documents and refine results step by step. Our article Five agents, one orchestrator describes the architecture in detail.
What daily work looks like
A new policy paper arrives. Instead of reviewing it by hand, a team member uploads it to ChatRPP. The agent analyses the content, highlights relevant passages, summarises the key points and classifies them according to RPP Group’s internal assessment logic. On request, it turns this into recommendations, narratives or lines of argument for different stakeholders.
| Task | Before | With ChatRPP |
|---|---|---|
| Document analysis | hours of manual review | automatic analysis, answers in real time |
| First drafts | by hand, often under time pressure | in seconds, in the consultancy’s tone |
| Quality assurance | by the team alone | QA agents check against internal standards |
| Team focus | routine crowds out strategy | strategic, client-facing work |
Responsibility and data protection
Responsible use of AI was a central principle of the project. ChatRPP supports political expertise in a targeted way: it takes on preparatory tasks, speeds up workflows and ensures consistency. Final assessment, classification and strategic decisions are always made by a person.
Political work is sensitive. All processing is therefore GDPR-compliant and runs in RPP Group’s own infrastructure, and confidential information never leaves the controlled environment.
The result
In practice: RPP Group
Time spent on manual routine work fell by around 70%. The team describes ChatRPP as an additional team member that understands the domain and delivers results needing little rework. Decisions are made faster because relevant information is available in real time.
The agent has become a fixed part of daily work, especially during periods of high information density and time pressure.
What other companies can learn from it
The project shows what AI agents can achieve beyond technical experiments when they are tailored to real workflows. Four points carry over to other knowledge-intensive industries:
- Start at the bottleneck: the agent targets the place where experienced people lose time to predictable routine.
- Mirror the way of working: assessment logic, tone and house standards belong in the system.
- Build in quality: reviewing agents and human approval keep results reliable.
- Keep data in-house: with confidential content, processing runs in your own infrastructure.
Our multi-agent development page shows how we build such systems for other processes.
An AI agent for your knowledge work?
In the process analysis we find the workflow where your team loses the most time to routine and show what an agent system for it could look like.
Request a process analysisFrequently asked questions
ChatRPP analyses policy documents, summarises them, derives strategic options and produces key messages and communication drafts in RPP Group’s style. Users can ask follow-up questions and refine results.
Time spent on manual routine work fell by around 70%. Answers are available in real time, and first drafts are produced in seconds.
Final assessment, classification and strategic decisions stay with people. The agent handles preparatory tasks.
In RPP Group’s own infrastructure and in a GDPR-compliant way. Confidential information never leaves the controlled environment.
Julia has been part of theBlue.ai since 2019 and has accompanied the development of AI applications in the enterprise environment since the company’s early days. In her role as Marketing Lead, she works closely with the engineering and consulting teams and makes complex technical topics understandable and accessible for decision-makers.
In her articles, she writes about practical experience from enterprise AI projects, as well as the challenges and opportunities of using AI in companies.