Dasseti’s embedded AI assistant, Sidekick, is about to redefine how our clients approach data, documentation, and decision-making, from due diligence to client engagement and ESG analysis.
Ahead of its full platform release in July, we spoke with Dasseti’s Head of Product, Arjun Patel, to unpack the design principles behind Sidekick and why this tool goes far beyond standard generative AI integrations.
A native, task-aware AI system
“Sidekick isn’t a bolt-on or third-party plugin,” explains Arjun. “It’s embedded directly into Dasseti’s core infrastructure, integrated with existing workflows, user permissions, and datasets.”
Its primary role: to automate the time-consuming, manual processes that our clients rely on daily, extracting structured data from unstructured documents, identifying inconsistencies, preparing and reviewing content, and enforcing quality control standards at scale.
Unlike generic LLM tools, Sidekick is built for institutional workflows, from RFP completion and policy reviews to manager assessments and risk reporting.
Designed for high-friction processes
Dasseti built Sidekick in response to concrete operational challenges seen across its client base:
- Due diligence analysts wasting hours on repetitive data entry, document parsing, and manual synthesis.
- RFP and investor relations teams struggling to maintain consistency across hundreds of documents and contributor inputs.
- ESG and operations teams needing tools that can extract, organise and summarise data without compromising on control or rigour.
“We focused on immediate-value use cases that required no behavioural change,” says Arjun. “Semantic search, document querying, and content generation and refinement were obvious starting points and are already delivering results.”
Trust, transparency and control
Enterprise-grade AI adoption hinges on trust and Sidekick was engineered with that principle at its core. Features include:
- Source referencing and confidence scoring across all outputs.
- Modular AI deployment, allowing clients to choose the features they want to use for AI.
- Zero data retention, ensuring alignment with strict data governance policies.
- Model flexibility, supporting a multi-modal infrastructure for LLM switching and hybrid deployment.
“Our users are managing billions (maybe trillions) in assets,” Arjun adds. “They need to know where every output comes from, and why. We designed Sidekick to make referencing and reasoning a non-negotiable.”
Future state: From tasks to workflows
While Sidekick already supports semantic search, iterative content drafting, document extraction and content review, the next evolution is task orchestration, allowing Sidekick to execute multi-step workflows.
“Soon, you’ll be able to interact directly via a chat interface, for example: ‘Prepare a quarterly risk report for all GPs in my coverage universe. Sidekick will gather the relevant data, review it against proprietary or customisable client assessment frameworks, generate insights, and provide a white-labelled, flexible output.
Of course, source referencing and reasoning will remain at the core.”
Other roadmap items include:
- Predictive analysis and trend detection across peer group data.
- Auto categorisation – automatically flag, tag and group your content.
- AI powered reporting and analytics.
- Tighter integrations with external tools & datasets (e.g. Microsoft, CRM tools, ADV databases).
Built for finance. Not retrofitted.
Sidekick’s differentiator lies in domain specialisation and system integration. “General-purpose AI tools are powerful, but they don’t understand our clients workflows,” Arjun says. “They introduce context-switching, fragmented outputs, and compliance risks. Sidekick operates within your system, on your data, with full traceability.”
The feedback so far? Strong adoption, fast onboarding, and high trust, even among initially hesitant clients. “Transparency has played a huge part in winning them over,” Arjun explains. “They can see the source references, confidence scores, even the reasoning behind every suggestion, and it becomes an indispensable tool.”
Sidekick will roll out to all Dasseti platform users in July.
For early access or a custom demo, get in touch.