TextForense

TextForense is a forensic textual communication analysis system that takes a text, an interview transcript, or an audio/video recording and, within minutes, returns a structured technical report: a Truth Score, linguistic stress indicators, behavioral patterns, psychological defense mechanisms, a Validity Analysis based on the scientific CBCA/SVA protocol, a Social Engineering investigation, a Fake News investigation, and an interpretive investigative analysis with suggested counterpoint questions.

Unlike a generic conversational AI tool adapted for any use case, TextForense was designed from the ground up exclusively for the task of textual investigation — which is what defines its technology classification.

TextForense is classified as a Specialized Application Artificial Intelligence for Textual Investigation — a category distinct both from general-purpose language models (which answer any question but were not designed for any specific investigative task) and from standalone fact-checking or sentiment-analysis tools (which cover only a fraction of the problem).

This specialization is what allows TextForense to combine, in a single analysis, forensic linguistics, applied behavioral psychology, credibility-assessment protocols historically used by forensic examiners, and factual verification — something that would otherwise require several standalone tools and specialized professionals working in parallel.

TextForense was developed in the United States by a multicultural team of specialists in investigation, forensic linguistics, and data science, with a track record — kept confidential given the sensitive nature of such projects — working alongside international intelligence agencies.

The investigative techniques built into the system are based on established scientific methods used by various intelligence and investigative bodies around the world, adapted and encoded into a platform accessible to any investigator, without requiring years of specialized training in forensic linguistics or testimony psychology.

In Brazil, the platform's operation and commercialization are carried out by Brasileiro Sem Fronteiras LLC, a company dedicated to the internationalization of individuals and assets, focused on helping Brazilians strategically legalize their lives and businesses abroad — ensuring support, reach, and operational continuity for Brazilian clients.

TextForense's analysis engine combines multiple recognized disciplines and protocols, without relying on a single isolated technique:

• CBCA (Criteria-Based Content Analysis), part of Statement Validity Assessment (SVA) — a protocol with decades of use in forensic and investigative contexts, applied automatically to the criteria that can be verified directly from the text.
• Forensic linguistics and natural language processing (NLP) to identify structure, coherence, and syntactic patterns associated with spontaneous versus rehearsed accounts.
• Applied behavioral psychology, including models such as Graham's Hierarchy, Cialdini's Principles of Influence, ego defense mechanisms, and catalogs of stratagems and logical fallacies.
• OSINT (Open Source Intelligence) and named entity recognition (NER) to check factual claims against open sources.
• Proprietary algorithms, developed in-house by cross-cultural investigation specialists, that interpret the linguistic signals extracted by the layers above and convert them into the percentages, classifications, and indices shown in each report.

Because they are company intellectual property, the exact parameters, weights, and computational instructions behind these proprietary algorithms are not publicly disclosed — which is itself part of why TextForense's output is difficult to reproduce with generic AI tools.

TextForense is based on scientific criteria and independent sources, never on opinion. In the Fake News module, for example, a claim about politics or the electoral process is checked against official sources such as Brazil's TSE (Superior Electoral Court) — because they are, by definition, the institutionally responsible source for that information.

This holds even when an institution is the target of public criticism or controversy: a scientific methodology references the officially responsible source for a topic, regardless of any value judgment about that institution's performance. Citing an official source does not mean TextForense agrees with, endorses, or defends that institution — it only means methodological rigor requires that reference, leaving it to the reader of the report to form their own judgment about the source's content.

The same principle — referencing established technical criteria, never opinion — governs the Narratives and Social Engineering modules: both apply linguistic analysis and narrative-structure techniques already documented in the literature (see section 4), identifying the presence or absence of specific patterns without making a value judgment about the reported content.

• Specialization, not generalization: every layer of the system was built and calibrated specifically for textual investigation, not adapted from a general-purpose model.
• Multidisciplinary combination: no standalone tool on the market brings together CBCA/SVA, forensic linguistics, social engineering, fake news verification, and psychological defense mechanisms in a single coherent report.
• Calibration driven by textual patterns: detection criteria are generalized from real linguistic patterns, not pinned to specific sentences — which reduces false positives and negatives as the system is tested against new accounts.
• Unpublished proprietary algorithms that translate complex linguistic signals into objective, comparable indices across analyses.
• Origin in high-technical-rigor intelligence projects, with investigative know-how that is not normally available outside government bodies or specialized agencies.

This combination of factors is what makes TextForense's output difficult to replicate with generic AI tools, manual checklists, or professionals working in isolation.

TextForense was designed to fit into the real workflow of investigators, without requiring extensive training:

• Paste, type, or transcribe (via audio/video upload or live recording) the statement, interview, or text to be analyzed.
• Select the investigation type — Narratives, Social Engineering, and/or Fake News — according to the case's objective.
• Review the generated report: Truth Score, alert timeline, Validity Analysis (CBCA), behavior, stress, defense mechanisms, and suggested investigative questions, among other layers.
• Use the report as technical supporting evidence, a script for new questions, or a basis for comparing multiple statements within the same case.
• Export the complete report as PDF or Word, with technical-responsibility fields (RACI) ready for signature, documentation, and filing.

• Time savings: what would take hours of manual reading, re-reading, and note-taking is delivered in minutes, with exact excerpts cited as evidence.
• Standardization: every statement goes through the same criteria and the same analysis layers, reducing variation in judgment between different analysts.
• Cross-case benchmarking: comparable indices (Truth Score, IEES, stress) let you compare statements against each other and over time, within the same case or across different cases.
• Ready-made question scripts: investigative questions automatically generated from the inconsistencies found, ready for the next interview.
• Defensible documentation: exportable reports, with cited methodology and evidence excerpts, ready to be part of a case file or to support the analysis of a qualified professional.
• Lower cost per case: replaces most of the manual initial-screening work, freeing the investigator to focus on the points that truly require human judgment.

TextForense serves professionals and organizations that need to assess the credibility of an account as part of their work, including:

• Private and corporate investigators (fraud, compliance, due diligence, background checks);
• Human Resources and Integrity/Compliance departments, in misconduct investigations and exit interviews;
• Lawyers and law firms, in preparing statements and conducting preliminary witness analysis;
• Insurance companies, in analyzing claim statements;
• Journalists and fact-checking teams;
• Public security forces and investigative bodies, as a technical support tool — never as a substitute for official forensic examination;
• Corporate security and competitive intelligence teams.

For anyone evaluating TextForense as part of a procurement process, the points below summarize the business case:

• Specialized technology, not an adapted general-purpose language model — with citable methodology (CBCA/SVA, forensic linguistics, OSINT) and its own proprietary algorithms.
• Origin and track record in projects with international intelligence agencies, with local operation in Brazil through Brasileiro Sem Fronteiras, ensuring support and operational continuity.
• Data handling aligned with LGPD, GDPR, and other international data protection laws — detailed in our Privacy Policy and Cookie Policy.
• The report is a technical support tool for human decision-making, not a forensic expert opinion — reducing the risk of relying on a single automated decision.
• Scalable plans, from individual use to intensive corporate use, with no need for a long implementation contract or complex integration.

To start an evaluation, create a free account and run your first analysis in minutes.

To talk to our sales team, visit the Contact page.