Impact Intern (Hiring Immediately)

Full-time - reports to Head of Impact
3 months fixed term contract (potential to extend)
2 days a week in London office, Euston (Tuesday & Thursday)
Salary £28,000 (Mac provided)

The Provenance Mission

Provenance is on a mission to empower 1 billion citizens to choose products that match their values, and we're building the tech to make it happen.

Our trusted platform enables transparency on the social and environmental impact of consumer products, and we’re looking for a mission-aligned intern to work closely with our Head of Impact in a support role.

Used by 300+ brands and retailers globally (including Holland & Barrett, Ocado, The Hut Group, MECCA) we are the data layer they need to become discoverable and credible to shoppers, LLMs, and AI agents reshaping how consumers buy. The shift to agentic commerce is happening now and we're right at the centre of it.

The Role Purpose

Validate product claims against the Provenance Data Standard and turn that expert judgement into the structured rules, schemas and evaluation sets that power Provenance's AI validation engine. The role combines hands-on claim review (initial focus) with the data structuring work that allows validation to scale, ensuring every claim surfaced to consumers and LLMs is substantiated, current and compliant.

1. Automating the Validation Engine

• Help translate proof point criteria into structured data point field references and deterministic rules, populating the validation logic and flagging where the criteria themselves need sharpening.
• Iterate on validation logic as patterns emerge, for example, moving from boolean to data point capture to get richer evidence from brands, or simplifying to binary where a rule is too ambiguous to apply consistently.
• Building the evaluation layer by curating gold-standard sets of approved and rejected claims with reasoning, so engine output can be tested against expert-validated ground truth as rules and models evolve.
• Schema evolution, identifying where current data structures fail and proposing structural updates.

2. Framework Evolution

• Keep Proof Point criteria current as certification standards, regulation and market practice evolve, so they stay defensible against the latest regulatory bar.
• Support the development of new proof point criteria and validation methodologies for new claim categories Provenance is expanding into. Research potential partnerships that can support its delivery
• Help grow the network of certifying partners whose standards Provenance recognises, identifying credible schemes, and onboarding them into the platform.
• Track regulatory developments across jurisdictions (ECGT, DMCC Act, CMA Green Claims Code, US state-level legislation, ESPR/DPP) and translate implications into concrete updates to the Data Standard, validation rules and customer-facing guidance.

3. Claims validation and dataset integrity

• Responsible for handling the claim approval queue within the appointed timelines, reviewing claims, giving structured feedback to customers and escalating cases internally where necessary.
• Keep Verified claims current: monitor claims expiry across the dataset, run the renewal process with brands, and work with product and engineering to automate detection and renewal wherever possible.
• First Customer Success point of call for complex claims across categories, including conducting further research where required, presenting summarised findings back to the team and feed them back into the Data Standard where new patterns emerge.
• Work closely with Customer Success Manager on onboarding
• Write and maintain the concise, factually precise, and regulator-proof one-liner templates that sit behind each proof point schemes

Requirements

Background: Final-year undergraduate or recent graduate in environmental science, sustainability, public policy, or a related field. A data-oriented background (environmental data science, geography with GIS, economics) would be a strong differentiator but is not essential, what matters is comfort working with structured information and an instinct for rigour over vagueness.

• Structured thinking and the ability to translate qualitative criteria (is this claim misleading?) into explicit, repeatable rules
• Interest in how AI systems consume and surface product information
• Attention to detail and consistency
• Ability to read and interpret certification standards, regulatory text, and technical documentation
• Clear, concise written communication
• Comfortable with LLM tooling, using LLMs to draft rules, test edge cases, spot inconsistencies